The Category P Features, Projections, Interpretation

The Category P Features, Projections, Interpretation The Category P Features, Projections, Interpretation

12.09.2014 Views

9 The problematic case for the non-uniform approaches is the locative construction (14) (this is acknowledged in Van Riemsdijk 1998). (14) Lisa put the book in the drawer. The locative PP in this construction is assumed to be an argument of the locative verb, rather than an adjunct (cf. Van Riemsdijk 1998, among others). Consequently, the PP in (14) has to receive a theta-role from the verb. This, however, creates a theory-internal paradox: Only PPs headed by a functional P can receive a theta-role. However, the P in (14) is a locative, semantically contentful P, which should be classified as ‘semi-lexical’, rather than as functional, and project a predicative PP, which is not assigned a theta-role. That the locative P in (14) has the same ‘semi-lexical’ status assumed for the locative P in (12) is supported by the binding paradigm (8a,b), which shows that locative PPs, regardless of their relation to the verb (i.e. argument vs. adjunct) behave uniformly. To conclude, the non-uniform approach makes an important step forward in not viewing P as a major lexical category, and in allowing at least some of its members to be functional. Furthermore, since diversity seems to be strongly associated with P in several dimensions, making an explicit assumption about it (in one of the dimensions), instead of ignoring it, is methodologically the right move. The specific dimension chosen in this approach (i.e. the lexical/functional dimension) is costly, as it results in a non-uniform theory of P which despite its non-uniformity is not sufficient to account for the array of challenging phenomena exhibited by P. Finally, if the lexical/functional distinction is a primitive option for individual Ps, then the more interesting question of why some Ps can be both ‘semi-lexical’ and functional (e.g. in, on), whereas others cannot (e.g. under, above), cannot be addressed, in principle. In light of the above, let us turn to examine the approach that takes P to be a uniformly functional head on a par with T, C or D. 1.1.2.2 The functional approach: Grimshaw (1991) develops a theory of Extended Projection, the goal of which is to derive the observation that functional categories such as T, C, D tend to appear with fixed complements, whereas lexical

10 heads such as N, V, A do not. 10 Under her approach a functional head functions as an extension of the appropriate lexical head projection (e.g. [DP-D’-D-NP-N’-N], [IP-I’- I-VP-V’-V]). It is proposed that P extends the nominal projection (i.e. PP-P’-P- DP…), similarly to C which is proposed to extend the verbal one. Since the principle which restricts the formation of an extended projection is that all heads in the extended projection are categorially identical, the categorial features of P in Grimshaw’s theory are that of N, [+N-V], rather than the [-N-V] cluster assumed in previous versions of the theory. The classification of P as a functional head (in the extended nominal projection) entails that P is not a theta-assigner. 11 The first apparent obstacle to Grimshaw’s uniform and restrictive theory is presented by semantically contentful Ps such as after, which can introduce clausal categories, arguably CPs. Since P is assumed to be part of the nominal extended projection, it cannot form an extended projection with a CP, which is verbal (i.e. part of the extended verbal projection). Since Ps are claimed to be functional, they are not supposed to either c(ategorially)- or s(emantically)-select their complement. To solve this problem Grimshaw proposes a certain relaxation regarding P. As opposed to the other functional heads, which do not s- or c-select, since their participation in the extended projection is guided by the principles of projection, semantically contentful Ps (referred to as ‘semantic’) are suggested to be allowed to s-select their complement. In this respect, note that although Grimshaw does not deny that some Ps are semantically contentful, she assumes that this does not necessarily preclude their being a functional category syntactically (Grimshaw 1991:7). However, even with the relaxation mentioned above, the wide distribution of PPs presents some prominent problems for Grimshaw’s theory. Recall that Ps can introduce not only argumental constituents, namely CPs and DPs, but also various predicative ones ((6b) repeated as (15b) and (15a,c)). This is virtually impossible under Grimshaw’s approach, as it is the basic premise of her theory that complements of the functional categories are fixed (note that the Ps in (15) are not semantically contentful, therefore s-selection cannot play any role here). 10 The same observation is made independently in Van Riemsdijk (1990). 11 See also Webelhuth (1992), where the claim that Ps function as affixes is taken to entail that P is not a theta-assigner.

9<br />

<strong>The</strong> problematic case for the non-uniform approaches is the locative<br />

construction (14) (this is acknowledged in Van Riemsdijk 1998).<br />

(14) Lisa put the book in the drawer.<br />

<strong>The</strong> locative PP in this construction is assumed to be an argument of the locative<br />

verb, rather than an adjunct (cf. Van Riemsdijk 1998, among others). Consequently,<br />

the PP in (14) has to receive a theta-role from the verb. This, however, creates a<br />

theory-internal paradox: Only PPs headed by a functional P can receive a theta-role.<br />

However, the P in (14) is a locative, semantically contentful P, which should be<br />

classified as ‘semi-lexical’, rather than as functional, and project a predicative PP,<br />

which is not assigned a theta-role.<br />

That the locative P in (14) has the same ‘semi-lexical’ status assumed for the<br />

locative P in (12) is supported by the binding paradigm (8a,b), which shows that<br />

locative PPs, regardless of their relation to the verb (i.e. argument vs. adjunct) behave<br />

uniformly.<br />

To conclude, the non-uniform approach makes an important step forward in not<br />

viewing P as a major lexical category, and in allowing at least some of its members to<br />

be functional. Furthermore, since diversity seems to be strongly associated with P in<br />

several dimensions, making an explicit assumption about it (in one of the<br />

dimensions), instead of ignoring it, is methodologically the right move. <strong>The</strong> specific<br />

dimension chosen in this approach (i.e. the lexical/functional dimension) is costly, as<br />

it results in a non-uniform theory of P which despite its non-uniformity is not<br />

sufficient to account for the array of challenging phenomena exhibited by P. Finally,<br />

if the lexical/functional distinction is a primitive option for individual Ps, then the<br />

more interesting question of why some Ps can be both ‘semi-lexical’ and functional<br />

(e.g. in, on), whereas others cannot (e.g. under, above), cannot be addressed, in<br />

principle.<br />

In light of the above, let us turn to examine the approach that takes P to be a<br />

uniformly functional head on a par with T, C or D.<br />

1.1.2.2 <strong>The</strong> functional approach: Grimshaw (1991) develops a theory of<br />

Extended Projection, the goal of which is to derive the observation that functional<br />

categories such as T, C, D tend to appear with fixed complements, whereas lexical

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!